US9733703B2 - System and method for on-axis eye gaze tracking - Google Patents

System and method for on-axis eye gaze tracking Download PDF

Info

Publication number
US9733703B2
US9733703B2 US14/626,288 US201514626288A US9733703B2 US 9733703 B2 US9733703 B2 US 9733703B2 US 201514626288 A US201514626288 A US 201514626288A US 9733703 B2 US9733703 B2 US 9733703B2
Authority
US
United States
Prior art keywords
eye
image
illumination
pupil
gaze
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active, expires
Application number
US14/626,288
Other languages
English (en)
Other versions
US20150160726A1 (en
Inventor
Nicholas J. SULLIVAN
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Mirametrix Inc
Original Assignee
Mirametrix Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Mirametrix Inc filed Critical Mirametrix Inc
Priority to US14/626,288 priority Critical patent/US9733703B2/en
Assigned to MIRAMETRIX INC. reassignment MIRAMETRIX INC. ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: TANDEMLAUNCH TECHNOLOGIES INC.
Assigned to TANDEMLAUNCH TECHNOLOGIES INC. reassignment TANDEMLAUNCH TECHNOLOGIES INC. ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: SULLIVAN, NICHOLAS JOSEPH
Publication of US20150160726A1 publication Critical patent/US20150160726A1/en
Application granted granted Critical
Publication of US9733703B2 publication Critical patent/US9733703B2/en
Assigned to WELLS FARGO BANK, NATIONAL ASSOCIATION, AS ADMINISTRATIVE AGENT reassignment WELLS FARGO BANK, NATIONAL ASSOCIATION, AS ADMINISTRATIVE AGENT SECURITY INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: LATTICE SEMICONDUCTOR CORPORATION
Active legal-status Critical Current
Adjusted expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/01Input arrangements or combined input and output arrangements for interaction between user and computer
    • G06F3/011Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
    • G06F3/013Eye tracking input arrangements
    • GPHYSICS
    • G02OPTICS
    • G02BOPTICAL ELEMENTS, SYSTEMS OR APPARATUS
    • G02B27/00Optical systems or apparatus not provided for by any of the groups G02B1/00 - G02B26/00, G02B30/00
    • G02B27/0093Optical systems or apparatus not provided for by any of the groups G02B1/00 - G02B26/00, G02B30/00 with means for monitoring data relating to the user, e.g. head-tracking, eye-tracking
    • G06K9/00604
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/18Eye characteristics, e.g. of the iris
    • G06V40/19Sensors therefor
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B3/00Apparatus for testing the eyes; Instruments for examining the eyes
    • A61B3/10Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions
    • A61B3/113Objective types, i.e. instruments for examining the eyes independent of the patients' perceptions or reactions for determining or recording eye movement
    • H04N13/0484
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/30Image reproducers
    • H04N13/366Image reproducers using viewer tracking
    • H04N13/383Image reproducers using viewer tracking for tracking with gaze detection, i.e. detecting the lines of sight of the viewer's eyes

Definitions

  • the following relates to systems and methods for on-axis eye gaze tracking.
  • LOS line-of-sight
  • the extracted eye features are used in conjunction with a chosen model of the eye to estimate the optical axis of the eye.
  • This axis determines the angular position of the eye in space, and can be used in conjunction with the known divergence of a user's visual axis to estimate where the user is looking in space.
  • Such advantages can include: an intuitive link between the visual system of the eye and the resultant images in the brain; the speed of eye movement relative to moving a hand-operated interaction device (i.e. users typically look at the desired destination of a hand-operated device prior to moving the hand-operated device); and the possibility that eye-gaze tracking techniques may be used by severely disabled individuals, to name a few.
  • a number of other applications for eye-gaze tracking systems can include, for example: psychological and physiological research into the connection between eye movements and perceptual and/or cognitive processes; an analysis of driver awareness; research into the effectiveness of advertising and website layouts; and gaze contingent displays, to name a few.
  • many existing gaze tracking technologies have been known to employ systems that operate as follows, wherein provided image data is analyzed for finding eyes, the found eyes have particular features extracted, and the features are used to estimate a point of gaze on the screen.
  • these systems typically employ multiple illumination sources and one or more imaging systems in order to determine the user's POG.
  • These systems tend to contain both on-axis and off-axis illumination sources, using an image differencing method to detect and track eyes in the scene. From this, the pupil center and multiple corneal reflections from the illumination sources (also known as Purkinje images) are extracted as eye features and used to determine the POG of the user on the screen.
  • a method of performing eye gaze tracking comprising: optimizing illumination of a scene for a single on-axis imaging apparatus; capturing an image using the single on-axis imaging apparatus under the optimized illumination; and processing the captured image to perform a gaze estimation.
  • FIG. 1 is a schematic diagram of an environment in which a gaze tracking system is incorporated into an electronic device for tracking the gaze of a user;
  • FIG. 2 is an example of an on-axis equipped electronic device
  • FIG. 3 is an example of an on-and-off axis equipped electronic device
  • FIG. 4 is an example of a configuration for a gaze tracking system
  • FIG. 5 is a schematic illustration of an eye in an image
  • FIG. 6 is block diagram illustrating an example of a configuration for a gaze tracking system
  • FIG. 7A is a state diagram illustrating application of a dynamic illumination procedure
  • FIG. 7B is a state diagram illustrating application of a dynamic illumination with multiple users in the scene
  • FIG. 8 is a schematic diagram illustrating a procedure for finding a second pupil
  • FIG. 9 is a schematic diagram illustrating a procedure for idealizing pupils
  • FIG. 10 is a schematic diagram illustrating another procedure for idealizing pupils
  • FIG. 11 is a flow chart illustrating example computer executable operations that may be performed in a dynamic illumination procedure
  • FIG. 12 is a flow chart illustrating an example of a process for on-axis eye candidate detection
  • FIG. 13 illustrates the effect of on-axis thresholding applied within an on-axis eye candidate detection process using idealized illumination parameters
  • FIG. 14 illustrates a flow chart illustrating another example of a process for on-axis eye candidate detection
  • FIG. 15 illustrates an example of an on-axis image
  • FIG. 16 illustrates an application of image gradient magnitude to the on-axis images of FIG. 15 ;
  • FIG. 17 illustrates an application of thresholding to the on-axis image of FIG. 15 ;
  • FIG. 18 is a flow chart illustrating example computer executable operations that may be performed in an eye candidate filtering procedure
  • FIG. 19 illustrates an eye illuminated by the gaze tracking system with pupil-glint vector and a distance metric used to normalize the illustrated vectors
  • FIG. 20 illustrates an eye illuminated by the gaze tracking system with pupil-glint vector and a limbus distance metric used to normalize the illustrated vector
  • FIG. 21 is a flow chart illustrating example computer executable operations that may be performed in a gaze determination procedure.
  • FIG. 22 is a flow chart illustrating example computer executable operations that may be performed by the gaze tracking system in an example implementation.
  • a system that is configured for sensing and tracking eye-gaze characteristics and to use information obtained therefrom to estimate a point-of-gaze.
  • Such a system is particularly advantageous in configurations using a single on-axis sensor to capture eye-gaze data.
  • the following provides a system and computer executable instructions and operations to be implemented by such a system for performing dynamic illumination of a subject's eyes, on-axis candidate detection techniques, on-axis candidate filtering processes, and on-axis gaze determination techniques as discussed in greater detail below.
  • the on-axis eye candidate filtering and on-axis eye candidate detection techniques may be generally referred to herein as eye detection and tracking.
  • the on-axis gaze determination techniques may also be referred to herein as methods by which extracted features from an image are used to estimate a gaze on a screen.
  • the dynamic illumination may be considered a parallel process or technique that modifies parameters of the apparatus being used to perform eye tracking to achieve acceptable image quality for gaze tracking.
  • the techniques described herein may be incorporated into any suitable eye gaze tracking system.
  • the dynamic illumination and eye candidate filtering algorithms can be independently modularized and used in any eye gaze tracking system to improve reliability and potentially increase the range of users under which gaze tracking will be feasible and/or functional.
  • system and methods described herein provide various advantages.
  • the system and methods described herein can be employed to reduced sizing, increase modularization, and increase the percentage of users under which gaze tracking is feasible.
  • a significant increase in modularability i.e. an ability to modularize
  • a depth calculation method is independent of the particular setup used.
  • the system described herein can be interfaced with the camera module in a manner similar to interfacing with a standard webcam.
  • This is particularly advantageous when compared to off-axis configurations that require more complex set up and interfacing to be used between the illumination sources and camera module, e.g., for synchronization, etc.
  • the distance of the off-axis illumination sources often limits the range in which an eye gaze tracker can function, since the distance metric is achieved via triangulation. It may be noted that this problem has also been found for eye tracking systems using two cameras, since two camera systems use the distance between the cameras to triangulate the distance the user is from the system.
  • the following techniques can advantageously enable an increased amount of variability in eye parameters and therefore be applicable to a larger user base.
  • the dynamic illumination technique described herein enables the eye gaze system to scan a more complete range of permissible illumination settings in order to find a user, whereas previous systems have been found to have a fixed range thus limiting the potential base of users. It has also been found that the system described herein can be adapted to handle other use cases such as where a user is wearing eyeglasses, by applying the eye candidate filtering algorithm to accurately reject pupil-like objects.
  • FIG. 1 illustrates a gaze tracking environment 8 (i.e. any environment in which gaze tracking is performed) in which an on-axis gaze tracking system 10 is incorporated into an electronic device 12 .
  • the gaze tracking system 10 tracks the gaze 14 of one or more eyes 16 of a user 18 which is directed towards at least a component of the electronic device 12 (e.g., display screen).
  • the gaze tracking system 10 includes one or more imaging components (described for example below) having a field of view (FOV) 20 that allows for imaging the eye(s) 16 of the user 18 .
  • FOV field of view
  • FIG. 2 illustrates an external front view of the on-axis equipped electronic device 12 shown in FIG. 2 .
  • the on-axis gaze tracking system 10 includes an imaging apparatus 22 that provides illumination and at least one lens element (not shown) for capturing images within the FOV 20 .
  • the on-and-off-axis equipped electronic device 12 ′ shown in FIG. 3 may include the on-axis gaze tracking system 10 herein described (with imaging apparatus 22 shown in FIG. 3 ), and includes additional off-axis imaging apparatus 24 (two apparatus 24 a , 24 b flanking on-axis apparatus 22 in FIG.
  • the on-axis gaze tracking system 10 is capable of being incorporated into smaller devices without requiring the additional separation for off-axis components. Moreover, the need for synchronization between components is greatly minimized if not eliminated.
  • FIG. 4 An example of a configuration for the on-axis gaze tracking system 10 is shown in FIG. 4 , which includes the on-axis imaging apparatus 22 and a processing module 30 .
  • the processing module 30 may generally represent any one or more components, modules, sets of programming instructions, etc. for performing the various techniques, processes, methods and algorithms (terms used interchangeably) described herein.
  • the processing module 30 is coupled to the imaging apparatus 22 to received captured images and to provide instructions and/or data for controlling use of the imaging apparatus 22 as described in greater detail below.
  • the on-axis imaging apparatus 22 includes an imaging module 34 (e.g., a camera), and one or more illumination sources 32 (e.g., a series of infrared LEDs surrounding a camera).
  • the on-axis gaze tracking system 10 also includes a communication bus or other connection 40 for interfacing with the electronic device 12 , e.g., to provide eye gaze as an input to an application or process on the electronic device 12 .
  • FIG. 5 is an example of an eye that may be imaged under either co-axial or on-axis illumination.
  • the eye includes a pupil 50 , an iris 52 , a glint 54 , sclera 56 , and surrounding skin 60 .
  • FIG. 6 an example of a configuration for the gaze tracking system 10 is shown.
  • various functional blocks of the processing module 30 are shown coupled to the imaging apparatus 22 .
  • the imaging apparatus 22 captures an on-axis image, which is provided to an image pre-processing block 72 for performing image enhancement techniques to the on-axis image.
  • image enhancement techniques can include any method of modifying the original image, to permit easier eye detection and tracking, or eye feature extraction. Examples of this could include de-noising, de-blurring, contrast enhancement, and edge enhancement of the image.
  • the pre-processing produces a modified image that is provided to an eye detection and tracking stage, in which eye candidate detection 74 is performed on the modified image to determine eye candidates, and eye candidate filtering 76 is performed on the eye candidates to determine “found” eyes.
  • the found eyes are provided to an eye feature extraction block 78 to determine eye features, which are provided to a gaze estimation block 80 to generate a point of gaze output, e.g., to be provided to an application or process in the electronic device 12 .
  • a dynamic illumination controller 70 can be used to obtain dynamic illumination parameters from the imaging apparatus 22 and eye feature intensity characteristics from the eye feature extraction block 78 to generate new dynamic illumination parameters to refine the illumination procedure on an on-going basis (e.g., to achieve an optimized or ideal illumination).
  • blob may be used in reference to a region of a digital image with common properties such that the region is considered a distinct object.
  • the digital image may be analyzed and filtered by a predetermined metric to be converted into a binary image (i.e. an image with pixels with a range of two values). In these cases, a binary blob is extracted.
  • a single illumination source may include any combination of smaller illumination sources such as LEDs to create a single unified illumination source, or a single large source.
  • the dynamic illumination process may function similar to an automatic gain control system, to ensure that the signal is within an acceptable illumination range, via a feedback-control loop in the system 10 .
  • the dynamic illumination process takes as an input one or more signal amplitude indicators at time step t, and one or more illumination variation parameters, and determines updated illumination parameters in order to improve the signal amplitude at time step (t+1).
  • the signal amplitude indicators for the dynamic illumination process in this example are intensity characteristics of the user's eyes, and as such can be deemed eye feature intensity parameters. Since one or more users may not always be contained within the scene, a multi-state system can also be configured to allow for such situations as shown in FIG. 7B .
  • the objective of the dynamic illumination controller 70 is to modify the effective scene illumination such that the user's eyes are at an optimized or “ideal” illumination.
  • the dynamic illumination controller 70 allows the gaze tracking system 10 to function accurately on a larger range of users, since there is a large amount of variation in users' pupil retro-reflectivity and having fixed camera parameters usually results in other gaze tracking systems not functioning accurately on a portion of the human population.
  • the intensity distribution of an extracted eye feature is used as the amplitude indicator.
  • the system assumes a relationship between a desired ideal illumination and a given signal amplitude indicator's values.
  • a dynamic illumination system could use the average intensity of a user's found pupils as its signal amplitude indicator, attempting to drive them as close as possible to half of the intensity range.
  • an illumination variation parameter can involve a parameter controlling the amount of illumination output by the on-axis illumination source during a given time step t, or one controlling the amount of illumination that is stored by the camera sensor.
  • An example of the former could be the amount of current passed to an LED illumination source; an example of the latter could be the shutter speed duration of an imaging system.
  • an on-axis image obtained from an on-axis apparatus may contain a user's pupils at an average intensity of 200 (as stored in an 8-bit unsigned greyscale image).
  • the average pupil brightness or pupil intensity of said user's eyes in this image would be 200.
  • average intensity we mean the average of the pixel values delineated as belonging to the user's pupils.
  • a coaxial illumination source refers to a source whose distance from the imaging system's optical axis is small enough that the reflected light returning to the system is substantially parallel to the axis of the optical system.
  • an ideal illumination can be considered an illumination setting where the desired eye features are illuminated enough such that they can be extracted by a computer vision system and are readily differentiable from other false positives in the scene.
  • an ideal illumination can be considered an illumination setting where the desired eye features are illuminated enough such that they can be extracted by a computer vision system and are readily differentiable from other false positives in the scene.
  • the principles of dynamic illumination may equally be applied to different configurations and different illumination criteria.
  • the ideal illumination concept follows from the differences in the reflective properties of the sclera, skin, pupil, pupil glint, and iris of a user.
  • iris 52 For all users under on-axis illumination, the order of reflectivity from lowest to highest is typically: iris 52 ⁇ sclera 56 ⁇ skin 60 ⁇ pupil 50 ⁇ corneal glint 54 (where ⁇ denotes an approximation, and ⁇ denotes lesser than—see also FIG. 5 ). It may be noted that although each user's pupil reflectivity properties may be quite different, the difference between pupil reflectivity and iris reflectivity is typically large enough in most if not all cases, that the relationships in reflectivity described herein should hold true.
  • a method for having a user's eyes at an on-axis ideal illumination can be described as follows: set the apparatus illumination properties such that the user's pupils are at a value corresponding to the midpoint of the range of the camera sensor. For example, for a camera sensor supplying an 8-bit digital image (0-255), the system's illumination settings can be modified such that the pupil is consistently at a value of 128. In this example, by doing so, the system 10 can aim to achieve the following:
  • the iris being the least reflective face property, being roughly within the lowest 1 ⁇ 4th of the image range and therefore easily distinguishable.
  • the dynamic illumination process herein described can take as input a scene's average pupil intensities as the signal amplitude parameter, and the current camera parameters, which in the described prototype uses shutter speed and gain. It may be noted that the process described herein may use camera parameters as the illumination variation parameters, as opposed to modifying the illumination source or other methods of modifying the scene's illumination. An objective of this process may therefore be to modify the effective scene illumination to approximate “ideal” conditions, and as such can be implemented via any combination of camera intensity parameters and illumination sources.
  • the dynamic illumination process can be described with the illumination variation parameters used being camera gain and camera shutter duration. As described earlier, the dynamic illumination system can consist of any 1 to n illumination variation parameters. Thus, it should be simple to imagine such a defined system consisting solely of the camera shutter duration or gain.
  • the dynamic illumination algorithm described herein may use the user's pupil intensity characteristics as the signal indicator parameters of the system 10 .
  • any number of other features could be used as the indicator parameters of this system 10 .
  • this feature is used in the described embodiment.
  • FIG. 7A illustrates the dynamic illumination process using a state diagram 90 having three states. Each state takes the pupils for analysis and determines whether or not to modify the camera intensity parameters. This example assumes that the system 10 expects only one user to be tracked in the scene at any given point in time.
  • the state diagram 90 can therefore be modified to a configuration for tracking a particular user when multiple users are in the scene, as shown in FIG. 7B
  • the dynamic illumination controller 70 cycles through different camera parameter settings until a pupil is found. In one example, the dynamic illumination controller 70 cycles through the full camera shutter duration range twice, switching the camera gain between 33% and 66% of its range at each cycle. A number of different shutter duration and gain combinations can be stored in a circular buffer and cycled through. A queue of found pupil intensities can also be created. At each iteration where a pupil is found, the pupil brightness is added into the queue.
  • the dynamic illumination process cycles through the circular buffer, changing the camera shutter duration to the new setting each time, until the queue is the size of the circular buffer, or no pupils are found and the queue is non-empty; that is, the pupil(s) have been found under a previously tested shutter duration.
  • the process examines the queue searching for the pupil brightness that is closest to the ideal pupil brightness, sets the shutter duration accordingly, and moves to the “Finding Other Pupil” state 92 shown in FIG. 7A .
  • the dynamic illumination process aims to choose the optimal shutter duration for the scene.
  • the objective is to locate the user's other pupil while avoiding losing track of the current pupil.
  • the dynamic illumination process can step up and down the camera's linear gain range until the second pupil is found, as depicted in FIG. 8 .
  • Points (a) and (b) in FIG. 8 delineate cases where the pupil is found to be at the edge of the desired pupil intensity range, and therefore a switch to searching in the other direction is performed, to avoid losing the pupil.
  • the second pupil is found and therefore the state is changed.
  • the dynamic illumination process may begin at the camera mid-gain value and proceed to step up the gain range until it reaches the upper limit, then returning to the mid-gain value, and stepping down the gain range until it reaches the lower limit.
  • the dynamic illumination controller 70 proceeds to do so until the second pupil is found, at which point the process transitions to a “Idealizing Pupils” state 96 as shown in FIG. 7A .
  • the dynamic illumination controller 70 may return to the searching for pupils state 94 . It may also be noted that if the scene is quite variable, a saw-tooth function can be implemented (as opposed to returning immediately to the midpoint of the range after reaching the top or bottom of the range). Moreover, the dynamic illumination process can be configured to increase or decrease the shutter duration once it has reached the limits of the gain range, to ensure it has cycled through the full range of the found pupil's accepted illumination.
  • an upwards or downwards motion stepping can be halted if the pupil's brightness is outside of set limits.
  • the limits used may include a sub range of the “Center Pupil Brightness” criteria used by the pupil candidate filtering process described below.
  • both pupils have been found and the objective in this state is to maintain or move the pupil intensity values as close as possible to the ideal pupil brightness to maintain the system in the ideal state, while ensuring both pupils maintain “in view” and are tracked consistently.
  • the process used herein can compare each pupil's brightness and determine which is furthest from the ideal pupil brightness.
  • the dynamic illumination controller 70 may then determine the direction needed to move the camera gain and/or shutter in order to make the pupil closer to the ideal.
  • the dynamic illumination controller can be configured to check that the other pupil is within a determined acceptable pupil range, for example, the same range described in the finding other pupil state 92 , described above. If so, the dynamic illumination controller 70 increments or decrements the camera gain by the gain step size. Otherwise, it maintains the current camera intensity parameters.
  • the idealizing pupil state 96 is illustrated in FIGS. 9 and 10 .
  • Pupil 2 is further from the ideal pupil brightness than Pupil 1, therefore the direction the whole setup needs to move in order to get it closer to the ideal brightness is determined. Since Pupil 1 is still within the acceptable pupil brightness range, the motion is allowed.
  • Pupil 1's brightness is furthest from the ideal, therefore the direction is determined based on that brightness. However, since Pupil 2 is outside the acceptable pupil brightness range, a camera intensity change does not need to be performed.
  • FIG. 7B describes a similar state diagram, where the “Searching for Pupil” state is replaced with a “Continuously Scanning for Pupils” state. Similar to the initial state in the single-user case, found pupils are stored, with their intensity and positions in the scene contained in a similar storage container. Distinguishable from the original state, however, this state does not need to switch to the “Finding other Pupil” state until it has received an external signal for the location of the desired user to be tracked.
  • the external signal may include a screen coordinate corresponding to the location of the desired user, as obtained from an external user location system.
  • the closest pupil to the screen point is determined and, if within an accepted distance threshold, the system switches to the “Finding other Pupil” state. Additionally, at any point the system may receive an external signal to switch tracked users. In such a case, the system returns to the initial state.
  • the described multi-user state diagram still only tracks a single user at a time, but permits selection of which user to track via an external signal.
  • the reason for this has been found to be that: unless the reflectivity properties of the users are similar, a dynamic illumination system 10 having one modifiable illumination source may only be able to track one user at a time.
  • the gaze tracking apparatus permits separate effective illumination switching of multiple sub-sections of the scene (via a complex set of illumination sources, or region-of-interest camera parameter switching, for example), this can easily be expanded to track multiple users. In such a situation, the system would include n sub-sections of the scene, each with their own 1 to m illumination variation parameters.
  • the described single-user state diagram could be implemented in each sub-section, allowing single user tracking within each sub-section of the scene.
  • the state diagram described in FIG. 7B can also be implemented for each sub-section.
  • Such a system would include n external control signals, and would permit multiple users per sub-section of the scene.
  • the dynamic illumination process performed by the dynamic illumination controller 70 can be described as follows.
  • the dynamic illumination controller 70 cycles through the dynamic illumination parameter range while eyes have not been found and determines at 202 whether or not an eye has been found. If an eye has been found, the dynamic illumination controller 70 can optionally finalize one pass through the dynamic illumination parameter range at 204 (as illustrated in dashed lines in FIG. 11 ), storing eye feature intensities for each available eye. The dynamic illumination controller 70 then determines at 206 whether there is a single user or multiple users in the scene.
  • the dynamic illumination controller 70 chooses the dynamic illumination parameter that best approximates the available users eyes to the ideal illumination at 208 or, if step 204 was skipped, proceeds straight to step 212 . If multiple users are expected in the scene, the dynamic illumination controller 70 determines the eyes to use based on the distance metric of the found eyes from an external indicator for the user of interest at 210 .
  • the dynamic illumination controller 70 dynamically modifies the illumination parameters at 212 such that the user's eyes remain as close as possible to the ideal illumination.
  • the metric used to maintain the eyes near the ideal illumination are the eye feature intensities in this example.
  • the example flow graph described herein does not contain a step associated with finding or re-finding one or multiple lost eyes as provided in at least some examples. Although this will result in a more robust tracking experience when both eyes are expected to be tracked, it has been found to not be required by the system. For example, a gaze tracking system focused on tracking single eyes and not eye pairs may not have an interest in perfecting the eye pair.
  • a common technique used by infrared-based eye gaze trackers for detecting a user's eyes involves thresholding the pupil 50 from the rest of the scene by first differencing the on-axis (bright) and off-axis (dark) images. Since the pupil 50 is bright due to the retro-reflective properties of the cornea in the on-axis image, and dark in the other, the resulting difference image should ideally just contain the pupils. This image would then be thresholded at a reliable level, resulting in the binary image containing the pupils.
  • additional issues such as noise caused by the differencing during user motion, and objects which show the same properties as the pupil, tend to add what can be a significant amount of false positives to the scene which needs to be filtered.
  • Other systems which use solely off-axis illumination, tend to perform fairly computationally expensive appearance-based methods of extracting and determining the eyes from the scene, such as template matching or multiple Haar-cascades classifiers.
  • the eye candidate detection 74 shown in FIG. 6 can be configured to advantage of the theorem that under on-axis ideal illumination conditions, the reflective properties of the user's iris 52 are noticeably lower than those of the rest of its face and its pupil 50 in the on-axis case, which studies have shown to be a reliable test set. It has also been found that studies on the reflective properties of human skin, the human iris, and the human pupil retina support these assumptions.
  • the user's pupils 50 can be easily extracted from the image. That is, using the knowledge that the contrast between the pupil and the iris 52 of a user's eyes will be one of the highest in the scene, binarizing the image to extract these objects can be made to be straightforward.
  • the blobs in such an image then only need to be analyzed and filtered (e.g., via the on-axis eye candidate filtering technique described below), leaving only the actual pupils 50 .
  • the determined threshold used should be such that under ideal illumination, any user can be thresholded accordingly.
  • the technique is dependent on the above-mentioned facets, namely that: a) the input image is under on-axis ideal illumination conditions, and b) the binary blobs can be correctly filtered, resulting in the actual pupils being selected. This can be particularly important, since the binary image may contain a large amount of noise, among them aspects of the user's face and potentially other noise in the scene.
  • Binarizing the image to extract binary blobs involves threhsolding the original image or a modified representation of the original image.
  • the advantage of implementing such a procedure using the described system is that the object of interest, the eye, is more easily differentiable from other objects in the scene due to the strong contrast between the pupil 50 and iris 52 .
  • two examples for creating a binary image are described below. It may be noted that these examples are provided only to demonstrate the advantageous effect of the provided prior knowledge for illustrative purposes, and should not be considered exhaustive within the principles described herein.
  • FIG. 12 illustrates one example of the eye candidate detection 74 .
  • a pre-processed image is obtained and the image is thresholded at 302 to generate a binary image.
  • the binary image is then processed to extract binary blobs at 304 and these binary blobs are ranked as pupil candidates at 306 .
  • the approach shown in FIG. 12 can be applied when the image scene corresponds primarily to the user's face, and illumination constraints are not a particular concern.
  • the system 10 may attempt to threshold the image at an intensity value approximating that of the iris 52 . By doing so, the skin 60 and pupil 50 should be segmented from the iris 52 , and it follows that the skin 60 and pupil 50 will be separate. Whether or not such segmentation includes the sclera 54 may be inconsequential. It has been found that thresholding at the bottom 1 ⁇ 4th of the illumination range of the camera sensor can be a reliable method of extracting the pupil 50 under such considerations.
  • FIG. 13 illustrates an example of the application of the process shown in FIG. 12 with on-axis thresholding within eye candidate detection under ideal illumination, where both exemplary binary images are potential results of the thresholding, whether the sclera 56 is linked with the face or not.
  • the end result in this example would be obtaining binary blobs corresponding to the actual pupils 50 and the whole or portions of the face, which would be filtered out at a later state.
  • FIG. 14 another example of a process for eye candidate detection 74 is shown.
  • a pre-processed image (e.g., image 350 shown in FIG. 15 ) is obtained at 320 and this image is used to calculate a gradient magnitude at 322 and a gradient orientation at 324 .
  • Calculating the gradient magnitude generates a gradient magnitude image as illustrated in FIG. 16 —image 360 , which undergoes gradient thresholding at 326 to obtain a binary image, e.g., as shown in FIG. 17 —image 370 .
  • the binary image is then processed at 328 to extract binary blobs, which are used with the gradient orientation to rank pupil candidates at 330 .
  • the process shown in FIG. 14 includes analyzing the image gradient of the input image, and determining an appropriate threshold for it.
  • This process takes into account imperfect illumination considerations, where additional illumination sources in the scene may make the simple thresholding above difficult to apply. Under such considerations, one may not be able to guarantee that the user's pupils will be able to reach the described ideal illumination in the scene. Also, the user's iris may not be within the expected intensity range from localized illumination making it brighter than expected. However, it can be assumed from prior knowledge of the scene that given the high retro-reflectivity properties of the human retina, the pupil should be one of the brightest objects in the scene. Additionally, one can presume that the iris 52 has a local minimal intensity. Thus, the gradient magnitude can be expected to be at the pupil-iris boundary one of the highest magnitudes in the scene.
  • a binarized image can be created which contains gradient magnitude binary blobs corresponding to the top P_threshold percentage of edges in the scene.
  • Both the proportion of the gradient signal corresponding to the eyes and that corresponding to other expected high contrast objects should use the maximum possible size and quantity of both of these, for the application described.
  • the system 10 can calculate the maximum expected size of the pupils (given the statistical variation in human pupil size and the operating range where the user is expected to be situated). For ideal binarization, a set proportion can be added, corresponding to the other expected high contrast objects (such as glasses glare), and threshold accordingly.
  • the system 10 is configured to select eyes in the binary image amongst what can be considerable amounts of noise.
  • FIG. 18 a two-step process may be employed to filter eyes from the binary image.
  • all eye candidates are filtered based on a set of criteria determined to best describe an eye and at 410 , the remaining eye candidates are analyzed for the “best fit” pair of eyes.
  • the method used to filter most eye candidates can be based on a machine learning methodology for creating reliable features, for example, considering that a large number of simple feature metrics tend to compound to provide a reliable and robust feature.
  • Each feature described is set with a given range within which an actual pupil is considered to acceptably lie under idealized on-axis illumination conditions.
  • Each eye candidate is then analyzed for all the features. If the eye candidate fails to lie within the range of one of them, it is removed from the candidate list.
  • a weighted feature set can be implemented, where suitable pupil candidates are those with a weighted sum of features above a certain threshold.
  • the system 10 is configured to consider average pupil intensity, center pupil intensity (a central cross-section of the pupil), and pupil intensity variance.
  • This metric allows small objects like simple noise, and large objects like the user's face to be easily filtered out.
  • the gradient magnitude of the pupil-iris boundary is expected that the gradient magnitude of the pupil-iris boundary to be within a certain range.
  • the system 10 can additionally ensure reliable pupil candidates by accepting candidates with an inward-directed gradient orientation. That is, candidates with a signal intensity that increases as the signal moves from the iris to the pupil are accepted.
  • the provided data can be analyzed and fit to ellipses corresponding to the pupil and/or iris ellipses. This fits with our expectation of the pupil and iris being circular objects rotated to some extent off of the image plane axis. Thus, the pupil and iris objects imaged on the sensor will correspond to some form of ellipse. Consequently, roundness, circularity, and intensity distributions can also be analyzed based on the ellipse estimate of the object.
  • this example analyzes only features pertaining to the pupil and iris of the user's eye for illustrative purposes.
  • a number of other features can be analyzed, such as the user's sclera, existence of corneal glints, and other facial features that would link a pupil candidate to a face.
  • all of these features could be combined into a common appearance-based matching approach, such as template matching or Haar-feature tracking.
  • the variability in pupil size relative to the iris, eye size due to its position in the scene, pupil reflectivity, and the use of glasses makes training and testing of such classifiers relatively difficult and computationally expensive.
  • the pupil candidates are considered for the best pair, to be deemed eyes. If only one pupil is present, the highest ranked one is chosen.
  • Finding the best pupil pair in this example includes comparing the described pupil features for each pair, attempting to find the most similar pair of eye candidates. This can be further enhanced by histogram matching of pupil candidates. In such an example, a successful pupil pair requires correlated histograms, as well as similar ellipse estimate dimensions and pupil intensity distributions. Additionally, knowledge of the inter-pupillary distance range between the pupils, as well as the potential size variability between them can further filter false positives.
  • the system 10 uses particular eye features to perform a gaze estimation 80 .
  • Typical Pupil Center Corneal Reflection (PCCR) mapping gaze estimation systems take one or more vectors ⁇ right arrow over (V i ) ⁇ , defined as the vector between the pupil center p c and a corneal glint g i , and using the one or more vectors, along with corresponding screen coordinates s i , create a gaze estimation mapping. This is performed via a calibration procedure which takes known pairs of features F and screen points s i , and performs a numerical fitting process to determine the coefficients of a functional mapping between them. Thus, a mapping function f(F) is created, where F is the set of features and f is the mapping function returning a screen point s i .
  • these gaze mappings have been found to deviate often significantly from the correct point-of-gaze when the user modifies his or her three-dimensional position.
  • one method when two corneal glints 54 are available, is to normalize the supplied vectors ⁇ right arrow over (V i ) ⁇ with a distance metric between the two corneal glints d g as shown in FIG. 19 .
  • FIG. 19 depicts an eye illuminated under such a system, along with the vectors and distance metric described.
  • the on-axis apparatus 22 since the on-axis apparatus 22 only creates a single corneal glint on each eye, using this standard normalized vector is not performed. Instead, a
  • d m is a distance metric that can correct for depth displacement.
  • the vector between the pupil center and corneal glint is determined at 500
  • the distance metric is determined at 502
  • the normalized vector is generated at 504 using the values obtained at 500 and 502 .
  • This normalized vector is then passed into a functional mapping f(F) at 506 , which outputs an estimated point-of-gaze on the screen.
  • Various distance metrics can be used to normalize the vector, including:
  • the distance between the center of the pupil and the iris/sclera boundary would be used.
  • Various approaches for extracting the limbus may be used, e.g., those used for performing biometric analysis via iris recognition.
  • This can be further simplified by extracting one point on the p limbus , given the pupil ellipse and corneal glint g i . It can be assumed that the pupil ellipse and limbus ellipse are concentric circles on a common plane, at an offset angle from the image plane they are imaged at. Thus, the major axis of each of the ellipses formed on the image is a reasonable approximation of the each circle's radius.
  • the relationship between the two radii can be found by calculating the relationship between the two intersection points i pupil and i limbus (from the pupil ellipse and limbus ellipse, respectively) on any line intersecting with the pupil center, p c . Therefore, given p limbus and the pupil ellipse, we extract the corresponding intersection point i pupil and determine the scalar relationship c by which the iris radius r iris is larger than the pupil radius r pupil . Finally, the distance metricd m , corresponding to the iris diameter, is calculated taking the pupil ellipse major axis a pupil (from the non-rotated ellipse equation
  • This metric assumes a model that can correct for cases where the user's head pose is not parallel to the camera, due to yaw motion, and in this example involves measurement of the distance between the user's pupils to determine distance from camera.
  • the correction for head pose could be solved by using a reliable other face feature that can be detected, such as a user's nose.
  • the distance metric is estimated based on the change in blur of particular objects in the scene.
  • the pupil/iris boundary is chosen, but other boundaries in the image (e.g. other eye or facial feature boundaries) can be chosen.
  • This metric can be obtained via either depth from focus or depth from defocus.
  • depth from focus the focus parameters are constantly switched such that the chosen object and object boundary are deemed in focus, with a focus criterion used to determine when the object has approximated perfect focus.
  • depth from defocus the object's blur is analysed directly at any point in time. In both cases, the returned criteria is used to estimate the depth of the object from the imaging system. This depth can easily be used as the distance metric for our normalized vector.
  • the system 10 may therefore be configured, in at least one example, to perform as shown in FIG. 22 .
  • an on-axis image is obtained, and a threshold image is generated at 602 .
  • the contours and blobs are then extracted at 604 from the thresholded image and the blobs are filtered for pupil candidates at 606 .
  • the dynamic intensity process is applied at 610 and camera intensity parameters 612 updated and utilized on an ongoing basis to achieve the ideal illumination as discussed above.
  • Pupil tracking is performed at 608 and in parallel, the dynamic intensity process is applied at 610 and camera intensity parameters 612 updated and utilized on an ongoing basis to achieve the ideal illumination as discussed above.
  • the corneal glint and limbus radius are extracted at 613 , and the described normalized feature vector is mapped to the location on the screen at 614 and the gaze on screen 616 is determined.
  • any module or component exemplified herein that executes instructions may include or otherwise have access to computer readable media such as storage media, computer storage media, or data storage devices (removable and/or non-removable) such as, for example, magnetic disks, optical disks, or tape.
  • Computer storage media may include volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information, such as computer readable instructions, data structures, program modules, or other data.
  • Examples of computer storage media include RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by an application, module, or both. Any such computer storage media may be part of the system 10 , any component of or related to the system 10 , etc., or accessible or connectable thereto. Any application or module herein described may be implemented using computer readable/executable instructions that may be stored or otherwise held by such computer readable media.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Optics & Photonics (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Ophthalmology & Optometry (AREA)
  • Multimedia (AREA)
  • Eye Examination Apparatus (AREA)
  • Studio Devices (AREA)
US14/626,288 2013-03-18 2015-02-19 System and method for on-axis eye gaze tracking Active 2034-07-25 US9733703B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US14/626,288 US9733703B2 (en) 2013-03-18 2015-02-19 System and method for on-axis eye gaze tracking

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US201361802881P 2013-03-18 2013-03-18
PCT/CA2014/050282 WO2014146199A1 (fr) 2013-03-18 2014-03-17 Système et procédé de suivi du regard sur axe
US14/626,288 US9733703B2 (en) 2013-03-18 2015-02-19 System and method for on-axis eye gaze tracking

Related Parent Applications (1)

Application Number Title Priority Date Filing Date
PCT/CA2014/050282 Continuation WO2014146199A1 (fr) 2013-03-18 2014-03-17 Système et procédé de suivi du regard sur axe

Publications (2)

Publication Number Publication Date
US20150160726A1 US20150160726A1 (en) 2015-06-11
US9733703B2 true US9733703B2 (en) 2017-08-15

Family

ID=51579241

Family Applications (1)

Application Number Title Priority Date Filing Date
US14/626,288 Active 2034-07-25 US9733703B2 (en) 2013-03-18 2015-02-19 System and method for on-axis eye gaze tracking

Country Status (6)

Country Link
US (1) US9733703B2 (fr)
EP (1) EP2975997B1 (fr)
JP (1) JP6577454B2 (fr)
CN (1) CN105431078B (fr)
CA (1) CA2882413C (fr)
WO (1) WO2014146199A1 (fr)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10061383B1 (en) * 2015-09-16 2018-08-28 Mirametrix Inc. Multi-feature gaze tracking system and method
US10984236B2 (en) 2015-12-31 2021-04-20 Mirametrix Inc. System and apparatus for gaze tracking
US11263634B2 (en) * 2019-08-16 2022-03-01 Advanced New Technologies Co., Ltd. Payment method and device

Families Citing this family (34)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP6504788B2 (ja) * 2014-11-12 2019-04-24 国立大学法人静岡大学 瞳孔検出用光源装置の制御方法
US10016130B2 (en) 2015-09-04 2018-07-10 University Of Massachusetts Eye tracker system and methods for detecting eye parameters
US10162651B1 (en) * 2016-02-18 2018-12-25 Board Of Trustees Of The University Of Alabama, For And On Behalf Of The University Of Alabama In Huntsville Systems and methods for providing gaze-based notifications
US10715824B2 (en) * 2016-03-17 2020-07-14 Facebook Technologies, Llc System and method for data compressing optical sensor data prior to transferring to a host system
EP3465536A1 (fr) 2016-05-27 2019-04-10 Jeff B. Pelz Système et procédé de suivi des yeux
US20180005057A1 (en) * 2016-07-01 2018-01-04 Hyundai Motor Company Apparatus and method for capturing face image of decreased reflection on spectacles in vehicle
JP6751324B2 (ja) * 2016-09-14 2020-09-02 株式会社デンソーアイティーラボラトリ 虹彩検出装置、虹彩検出方法、およびプログラム
US20180081434A1 (en) * 2016-09-22 2018-03-22 Apple Inc. Eye and Head Tracking
CN112578911A (zh) * 2016-12-06 2021-03-30 美国景书公司 跟踪头部和眼睛的运动的装置和方法
WO2018119077A1 (fr) 2016-12-21 2018-06-28 Acucela Inc. Système de tomographie par cohérence optique à faible coût, mobile et miniaturisé pour applications ophtalmiques à domicile
US10296786B2 (en) * 2017-02-15 2019-05-21 International Business Machines Corporation Detecting hand-eye coordination in real time by combining camera eye tracking and wearable sensing
WO2018222897A1 (fr) * 2017-06-01 2018-12-06 University Of Washington Pupillomètre numérique basé sur un téléphone intelligent
CN107357429B (zh) * 2017-07-10 2020-04-07 京东方科技集团股份有限公司 用于确定视线的方法、设备和计算机可读存储介质
JP7116978B2 (ja) * 2017-07-12 2022-08-12 国立大学法人静岡大学 画像特徴追跡方法、視線追跡方法及び注視点追跡方法
EP4011274A1 (fr) * 2017-09-08 2022-06-15 Tobii AB Suivi de l' il utilisant une position centrale du globe oculaire
EP4137037A1 (fr) * 2017-09-11 2023-02-22 Tobii AB Fiabilité de données de suivi du regard pour l' il gauche et droit
US11690509B2 (en) * 2017-12-13 2023-07-04 Medical Diagnostech Pty Ltd. System and method for obtaining a pupil response profile
JP7402866B2 (ja) 2018-06-20 2023-12-21 アキュセラ インコーポレイテッド 家庭用眼科用途のための小型化モバイル低コスト光干渉断層撮影システム
US11003244B2 (en) * 2018-08-27 2021-05-11 University Of Rochester System and method for real-time high-resolution eye-tracking
EP3844553A2 (fr) * 2018-08-31 2021-07-07 Eyeware Tech SA Procédé et système d'estimation du regard
CN109725715B (zh) * 2018-11-16 2024-04-05 深圳惠智光电子科技有限公司 一种眼睛训练引导方法、电子设备及存储介质
IL264530B1 (en) * 2019-01-29 2024-03-01 Eyeway Vision Ltd Device for tracking eye movement and method therefor
CN110051319A (zh) * 2019-04-23 2019-07-26 七鑫易维(深圳)科技有限公司 眼球追踪传感器的调节方法、装置、设备及存储介质
US10996753B1 (en) * 2020-04-07 2021-05-04 Eyetech Digital Systems, Inc. Multi-mode eye-tracking with independently operable illuminators
US11921917B2 (en) 2020-04-07 2024-03-05 Eyetech Digital Systems, Inc. Compact eye-tracking camera systems and methods
US11662810B2 (en) * 2020-06-05 2023-05-30 Magic Leap, Inc. Enhanced eye tracking techniques based on neural network analysis of images
WO2022035809A1 (fr) 2020-08-14 2022-02-17 Acucela Inc. Système et procédé d'alignement par courbure décroissante de balayage a de tomographie par cohérence optique
WO2022084803A1 (fr) 2020-10-23 2022-04-28 Eyeware Tech Sa Procédé d'étalonnage automatisé d'un système comprenant un dispositif externe de suivi oculaire et un dispositif informatique
EP4027220A1 (fr) 2021-01-08 2022-07-13 Eyeware Tech SA Procédé d'étalonnage d'un dispositif de suivi oculaire
EP4113982A1 (fr) 2021-06-30 2023-01-04 Eyeware Tech SA Procédé de détection et de communication du centre d'attention visuel dans une vidéoconférence
EP4113983A1 (fr) 2021-06-30 2023-01-04 Eyeware Tech SA Procédé d'étalonnage de suivi du regard avec un système de vidéoconférence
EP4113251A1 (fr) 2021-06-30 2023-01-04 Eyeware Tech SA Procédé d'étalonnage d'un système comprenant un dispositif de suivi de l' il et un dispositif informatique comprenant un ou plusieurs écrans
US11861916B2 (en) * 2021-10-05 2024-01-02 Yazaki Corporation Driver alertness monitoring system
CA3237333A1 (fr) * 2022-06-02 2023-12-07 Thomas Johannes WAWRZINEK Oculometre a filtrage materiel et tri logiciel

Citations (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US3804496A (en) 1972-12-11 1974-04-16 Stanford Research Inst Two dimensional eye tracker and method for tracking an eye
US4950069A (en) 1988-11-04 1990-08-21 University Of Virginia Eye movement detector with improved calibration and speed
US5231674A (en) * 1989-06-09 1993-07-27 Lc Technologies, Inc. Eye tracking method and apparatus
US6134339A (en) 1998-09-17 2000-10-17 Eastman Kodak Company Method and apparatus for determining the position of eyes and for correcting eye-defects in a captured frame
US6634749B1 (en) * 1998-11-02 2003-10-21 Leica Microsystems (Schweiz) Ag Eye tracking system
WO2007092512A2 (fr) 2006-02-07 2007-08-16 Attention Technologies, Inc. Dispositif de surveillance d'assoupissement et de distraction du conducteur
US7362885B2 (en) 2004-04-20 2008-04-22 Delphi Technologies, Inc. Object tracking and eye state identification method
US20100007601A1 (en) 2006-07-28 2010-01-14 Koninklijke Philips Electronics N.V. Gaze interaction for information display of gazed items
US20100057059A1 (en) 2006-11-29 2010-03-04 Nidek Co., Ltd. Corneal surgery apparatus
US7762665B2 (en) 2003-03-21 2010-07-27 Queen's University At Kingston Method and apparatus for communication between humans and devices
CN101951828A (zh) 2008-05-12 2011-01-19 丰田自动车株式会社 驾驶员成像装置以及驾驶员成像方法
US20110175932A1 (en) 2010-01-21 2011-07-21 Tobii Technology Ab Eye tracker based contextual action
US8077914B1 (en) * 2006-08-07 2011-12-13 Arkady Kaplan Optical tracking apparatus using six degrees of freedom
US8292433B2 (en) 2003-03-21 2012-10-23 Queen's University At Kingston Method and apparatus for communication between humans and devices
US8339446B2 (en) 2009-04-01 2012-12-25 Tobii Technology Ab Adaptive camera and illuminator eyetracker
US8342687B2 (en) 2009-10-08 2013-01-01 Tobii Technology Ab Eye-tracking using a GPU
US8343067B2 (en) 2005-02-24 2013-01-01 Warren Jones System and method for quantifying and mapping visual salience
US8348429B2 (en) * 2008-03-27 2013-01-08 Doheny Eye Institute Optical coherence tomography device, method, and system
CA2545202C (fr) 2003-11-14 2014-01-14 Queen's University At Kingston Procede et appareil de poursuite oculaire sans etalonnage

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2006023949A (ja) * 2004-07-07 2006-01-26 Canon Inc 画像処理装置及び画像処理方法
JP4780088B2 (ja) * 2007-11-09 2011-09-28 アイシン精機株式会社 顔画像撮像装置、顔画像撮像方法、及びそのプログラム

Patent Citations (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US3804496A (en) 1972-12-11 1974-04-16 Stanford Research Inst Two dimensional eye tracker and method for tracking an eye
US4950069A (en) 1988-11-04 1990-08-21 University Of Virginia Eye movement detector with improved calibration and speed
US5231674A (en) * 1989-06-09 1993-07-27 Lc Technologies, Inc. Eye tracking method and apparatus
US6134339A (en) 1998-09-17 2000-10-17 Eastman Kodak Company Method and apparatus for determining the position of eyes and for correcting eye-defects in a captured frame
US6634749B1 (en) * 1998-11-02 2003-10-21 Leica Microsystems (Schweiz) Ag Eye tracking system
US8096660B2 (en) 2003-03-21 2012-01-17 Queen's University At Kingston Method and apparatus for communication between humans and devices
US7762665B2 (en) 2003-03-21 2010-07-27 Queen's University At Kingston Method and apparatus for communication between humans and devices
US8322856B2 (en) 2003-03-21 2012-12-04 Queen's University At Kingston Method and apparatus for communication between humans and devices
US8292433B2 (en) 2003-03-21 2012-10-23 Queen's University At Kingston Method and apparatus for communication between humans and devices
CA2545202C (fr) 2003-11-14 2014-01-14 Queen's University At Kingston Procede et appareil de poursuite oculaire sans etalonnage
US7362885B2 (en) 2004-04-20 2008-04-22 Delphi Technologies, Inc. Object tracking and eye state identification method
US8343067B2 (en) 2005-02-24 2013-01-01 Warren Jones System and method for quantifying and mapping visual salience
WO2007092512A2 (fr) 2006-02-07 2007-08-16 Attention Technologies, Inc. Dispositif de surveillance d'assoupissement et de distraction du conducteur
US20100007601A1 (en) 2006-07-28 2010-01-14 Koninklijke Philips Electronics N.V. Gaze interaction for information display of gazed items
US8077914B1 (en) * 2006-08-07 2011-12-13 Arkady Kaplan Optical tracking apparatus using six degrees of freedom
US20100057059A1 (en) 2006-11-29 2010-03-04 Nidek Co., Ltd. Corneal surgery apparatus
US8348429B2 (en) * 2008-03-27 2013-01-08 Doheny Eye Institute Optical coherence tomography device, method, and system
CN101951828A (zh) 2008-05-12 2011-01-19 丰田自动车株式会社 驾驶员成像装置以及驾驶员成像方法
US8339446B2 (en) 2009-04-01 2012-12-25 Tobii Technology Ab Adaptive camera and illuminator eyetracker
US8342687B2 (en) 2009-10-08 2013-01-01 Tobii Technology Ab Eye-tracking using a GPU
US20110175932A1 (en) 2010-01-21 2011-07-21 Tobii Technology Ab Eye tracker based contextual action

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
English translation of search report from corresponding Chinese Application No. 201480028893.9, issued Jan. 18, 2017.
Norman, P.; International Search Report from corresponding PCT Application No. PCT/CA2014/050282; search completed Jun. 27, 2014.
Yoo, H. D. et al.; "A novel non-intrusive eye gaze estimation using cross-ration under large head motion"; Computer Vision and Image Understanding 98 (2005) 25-51; Elsevier; Accessed online Jun. 24, 2014 from http://www.shoutwiki.com/w/images/wearable/9/9f/Yoo-2005-Computer-Vision-and-Image-Understanding.pdf.
Yoo, H. D. et al.; "A novel non-intrusive eye gaze estimation using cross-ration under large head motion"; Computer Vision and Image Understanding 98 (2005) 25-51; Elsevier; Accessed online Jun. 24, 2014 from http://www.shoutwiki.com/w/images/wearable/9/9f/Yoo—2005—Computer-Vision-and-Image-Understanding.pdf.

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10061383B1 (en) * 2015-09-16 2018-08-28 Mirametrix Inc. Multi-feature gaze tracking system and method
US10984236B2 (en) 2015-12-31 2021-04-20 Mirametrix Inc. System and apparatus for gaze tracking
US11263634B2 (en) * 2019-08-16 2022-03-01 Advanced New Technologies Co., Ltd. Payment method and device

Also Published As

Publication number Publication date
CA2882413A1 (fr) 2014-09-25
JP2016512765A (ja) 2016-05-09
JP6577454B2 (ja) 2019-09-18
CA2882413C (fr) 2016-10-25
CN105431078A (zh) 2016-03-23
EP2975997B1 (fr) 2023-07-12
US20150160726A1 (en) 2015-06-11
EP2975997A1 (fr) 2016-01-27
WO2014146199A1 (fr) 2014-09-25
CN105431078B (zh) 2018-11-02
EP2975997A4 (fr) 2016-12-14

Similar Documents

Publication Publication Date Title
US9733703B2 (en) System and method for on-axis eye gaze tracking
US10692210B2 (en) Recording medium storing computer program for pupil detection, information processing apparatus, and pupil detecting method
US9411417B2 (en) Eye gaze tracking system and method
CN110266916B (zh) 用于在眼睛追踪中处理眩光的方法和系统
US11715231B2 (en) Head pose estimation from local eye region
Chen et al. Efficient and robust pupil size and blink estimation from near-field video sequences for human–machine interaction
US10311583B2 (en) Eye motion detection method, program, program storage medium, and eye motion detection device
WO2011158463A1 (fr) Dispositif d'évaluation de l'éblouissement provoqué par la lumière extérieure, dispositif de détection d'une ligne visuelle et procédé d'évaluation de l'éblouissement provoqué par la lumière extérieure
KR20120006819A (ko) 시선 추적 방법 및 이를 적용하는 장치
CN114391117A (zh) 眼睛跟踪延迟增强
US11163994B2 (en) Method and device for determining iris recognition image, terminal apparatus, and storage medium
KR102442220B1 (ko) 얼굴 생체 검출 방법과 장치, 전자 기기, 컴퓨터 판독 가능 매체
CN114092985A (zh) 一种终端控制方法、装置、终端及存储介质
Wan et al. Robust and accurate pupil detection for head-mounted eye tracking
Heo et al. Object recognition and selection method by gaze tracking and SURF algorithm
JP2004192552A (ja) 開閉眼判定装置
Fuhl et al. Pupil detection in the wild: An evaluation of the state of the art in mobile head-mounted eye tracking
CN113093907B (zh) 人机交互方法、系统、设备及存储介质
KR102074977B1 (ko) 전자 장치 및 그의 제어 방법
Kim et al. Eye detection for gaze tracker with near infrared illuminator
Ferhat et al. Eye-tracking with webcam-based setups: Implementation of a real-time system and an analysis of factors affecting performance
KR100311605B1 (ko) 컴퓨터비전기반의시선추적방법
KR20010001021A (ko) 눈 깜빡임을 이용한 졸음 감지 방법 및 졸음 레벨 검출 전처리과정에서의 변형되는 눈 영역 추적 방법
García-Dopico et al. Precise Non-Intrusive Real-Time Gaze Tracking System for Embedded Setups.
Park Real-time gaze detection via neural network

Legal Events

Date Code Title Description
AS Assignment

Owner name: MIRAMETRIX INC., CANADA

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:TANDEMLAUNCH TECHNOLOGIES INC.;REEL/FRAME:034986/0898

Effective date: 20140321

Owner name: TANDEMLAUNCH TECHNOLOGIES INC., CANADA

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:SULLIVAN, NICHOLAS JOSEPH;REEL/FRAME:034986/0670

Effective date: 20130626

STCF Information on status: patent grant

Free format text: PATENTED CASE

CC Certificate of correction
MAFP Maintenance fee payment

Free format text: PAYMENT OF MAINTENANCE FEE, 4TH YR, SMALL ENTITY (ORIGINAL EVENT CODE: M2551); ENTITY STATUS OF PATENT OWNER: SMALL ENTITY

Year of fee payment: 4

FEPP Fee payment procedure

Free format text: ENTITY STATUS SET TO UNDISCOUNTED (ORIGINAL EVENT CODE: BIG.); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

MAFP Maintenance fee payment

Free format text: PAYMENT OF MAINTENANCE FEE UNDER 1.28(C) (ORIGINAL EVENT CODE: M1559); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

FEPP Fee payment procedure

Free format text: PETITION RELATED TO MAINTENANCE FEES GRANTED (ORIGINAL EVENT CODE: PTGR); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

AS Assignment

Owner name: WELLS FARGO BANK, NATIONAL ASSOCIATION, AS ADMINISTRATIVE AGENT, COLORADO

Free format text: SECURITY INTEREST;ASSIGNOR:LATTICE SEMICONDUCTOR CORPORATION;REEL/FRAME:061365/0560

Effective date: 20220901